Not applicable.
Not applicable.
Not applicable.
The present invention relates to devices for optical signal routing, and more particularly to a non-mechanical solid-state optical switch.
In optical communication networks, light signals are transmitted along optical paths, such as optical fibers. Although the propagation of light signals along optical fibers is somewhat analogous to the transmission of electronic signals along wires, transferring the light signals from one optical fiber to another is somewhat more difficult than coupling electrical wires together. In particular, switching a light signal from one optical path to another, such as from an input fiber to one of several output fibers, has been more difficult than similar switching of electrical signals for a variety of reasons.
Generally, when switching an optical signal, the light from one optical fiber is coupled to another optical fiber. Several approaches have been used to achieve such optical switching. One approach is to physically move the end of one fiber between positions so that it becomes aligned with an end of the desired output fiber. However, modem optical fibers have a very small across section, and typically have a fairly narrow acceptance angle within which light entering the fiber must fall to promote efficient propagation of the light along the fiber. Therefore, such optic switches must not only be precisely aligned, but also must stay suitably aligned. Additionally, the fibers themselves are glass and must be handled carefully to avoid breakage, which may be exacerbated by the physical motion of the fiber each time switching occurs.
Other approaches hold the fibers stationary while moving an optical element, such as a mirror or prism, to direct the light signal from the input fiber to one of the output fibers. Such approaches are described in U.S. Pat. Nos. 5,838,847 and 5,959,756. While the problem of flexing the optical fibers is avoided by these approaches, they generally require precise alignment of the optical components, including the fiber ends, and should retain this alignment over the operating lifetime of the switch. Mechanical factors, such as wear and mechanical shock, can affect the alignment of the switch. Furthermore, such switches are relatively large, in light of the movable mechanical structure, which can also affect switching speed because of the inertia of the mechanics, and can consume a relatively large amount of electrical energy to switch, particularly at high switching speeds. Some mechanical switches are also susceptible to bouncing, which can lengthen the time required for reliable switching.
As the demand for optical communications expands, the number of switches required for the optical networks that carry the optical signals increases. One factor contributing to the increased number of switches is the growing use of wave-division multiplexing (xe2x80x9cWDMxe2x80x9d). In a WDM system, a single optical fiber carries many different xe2x80x9cchannelsxe2x80x9d, which are generally portions of the optical spectrum that can be separately routed and otherwise manipulated from each other. For example, a WDM optical network carrying four channels on a single fiber might service two users, each with two channels. Two of the channels can be routed (xe2x80x9cdroppedxe2x80x9d) to the first user, while the other two channels continue to the next user. In order to fully utilize all the channels, optical signals occupying the dropped channels can be added onto the optical fiber. Thus, it may be desirable provide the capability to switch each channel separately from the others.
As the number of optical channels carried on an optical fiber increases, the number of switches is likely to similarly increase. Some optical networks are using dense wave-division multiplexing (xe2x80x9cDWDMxe2x80x9d), with channels spaced every 100 GHz. Other networks are being planned where the channels are spaced even closer, and where more of the optical band is being used to carry channels. Optical communication systems might be developed that have a need for significantly more switches than current systems. It may be difficult to accommodate an expansion of the existing fiber network with sufficient space for the necessary additional switches.
Fabry-Perot based structures have been used in a variety of laser-switching functions, but the structures developed for these applications are generally unsuitable for optical telecommunication switching purposes. For example, among the most numerous Fabry-Perot structures are the asymmetric structures, in which the reflectivity of one cavity mirror differs from the other. Asymmetric configuration generally reduces the photon lifetime in the cavity, allowing for ultrafast (picosecond) switching. In addition the electric field is enhanced in the cavity, increasing the overall absorption. The majority of these devices consist of thin films of semiconductor materials.
The asymmetric Fabry-Perot devices are primarily used as reflection modulators. In the majority of configurations the incident light enters the device through the weaker of the two reflectors. When the spacer layer is in an absorbing state, the reflectivity of the device is low. When the spacer layer is in a transparent state the reflectivity is primarily from the back reflector and is of a high level. In many cases these devices consist of a single layer of high index material on a high reflector. A predominant method for altering the absorption is through an electric field induced shift in the semiconductor band edge that occurs either through the Franz-Keldysh effect (in bulk media) or the quantum confined Stark effect (in quantum well structures.)
These asymmetric Fabry-Perot devices are not suitable for reflection/transmission switches for the telecommunication market for several reasons. As these devices are operated in a reflectance mode only, the structures are typically grown on substrates that are absorbing at the wavelength of interest. In addition, the asymmetric mirrors often preclude low transmission loss.
Other asymmetric Fabry-Perot devices use all optical switching through nonlinear effects. Refractive index changes are generally used in these devices to change the spacer center-wavelength and hence the spectral position of the reflectance minimum. In an optical communication application, the shift in the frequency response of the reflectance could interfere with other optical channels on the system.
There are also a number of Fabry-Perot based devices whose spacer layers exhibit a change in refractive index, and hence effective optical path length with an external stimulus. These typically are intended to be tunable, since changing the refractive index of the spacer layer changes the resonant frequency of the cavity. A problem with these devices in switching applications is that they will sweep across other wavelengths and disrupt communication in those channels. In fact, most devices that modulate the index cannot stay in the same channel passband while changing state.
Another popular class of devices involves all-optical switching with a bistable resonator configuration, often with the spacer layer consisting of a saturable absorber. Planar semiconductor Fabry-Perot laser structures have been analyzed and demongrated for this application. Vertical-cavity semiconductor laser (xe2x80x9cVCSELxe2x80x9d) structures have also been demonstrated with an all-optical bi-stable switching behavior. These amplifiers have been pumped both optically and electrically. VCSEL structures are thin film based devices in which a quarter-wave stack mirrors and the spacer layer(s) are typically all semiconductor materials. One advantage of the VCSEL structure is compactness compared to a waveguide-based semiconductor amplifier device. However, these devices generally do not provide for transmission through them, as the substrate material that VCSELs are fabricated an is typically opaque at the wavelength of interest. In addition, VCSEIs are predominately used as lasers, which require a different set of design parameters (such as high gain) than optical switches and attenuators.
All-optical bistable switching has been researched for use in optical computing applications, where such a device would function as an xe2x80x9coptical transistorxe2x80x9d. Such optical bistable switches are not suitable for current telecommunications switching applications; however, because they are dependent on the intensity level of the incident signal. The spacer medium is nominally absorbing, so at low incident light levels the device is non-transmitting. If the intensity of the incident light is sufficiently increased, the absorbing mechanism in the spacer layer will saturate, and the device will transmit light. A high E-field builds up in the spacer layer once the device has become transmitting, and the spacer layer will remain transparent for lower incident light levels than that which initially turned on the device. Thus, the device is non-reciprocal, and would absorb weak optical signals, even if the switch were xe2x80x9conxe2x80x9d.
Thus, it is desirable to provide an optical switch that is relatively small in size, reliable in operation, with a long operating lifetime, and that consumes relatively little power to switch states, and that can switch a channel carried on an optical communication network without disrupting adjacent channels. In a particular application, the optical switch should have very low loss ( less than 0.5 dB) in either state and a high switching speed ( less than 10 ns) in order for efficient operation in all-optical networks.